Reliability, Repeatability, and Reproducibility of Pulmonary Transit Time Assessment by Contrast Enhanced Echocardiography Ingeborg H

Herold et al. Cardiovascular Ultrasound (2016) 14:1 DOI 10.1186/s12947-015-0044-1

RESEARCH Open Access Reliability, repeatability, and reproducibility of pulmonary transit time assessment by contrast enhanced echocardiography Ingeborg H. F. Herold1*, Salvatore Saporito2, R. Arthur Bouwman1, Patrick Houthuizen3, Hans C. van Assen2, Massimo Mischi2 and Hendrikus H. M. Korsten1,2

Abstract Background: The aim of this study is to investigate the inter and intra-rater reliability, repeatability, and reproducibility of pulmonary transit time (PTT) measurement in patients using contrast enhanced ultrasound (CEUS), as an indirect measure of preload and left ventricular function. Methods: Mean transit times (MTT) were measured by drawing a region of interest (ROI) in right and left cardiac ventricle in the CEUS loops. Acoustic intensity dilution curves were obtained from the ROIs. MTTs were calculated by applying model-based fitting on the dilution curves. PTT was calculated as the difference of the MTTs. Eight raters with different levels of experience measured the PTT (time moment 1) and repeated the measurement within a week (time moment 2). Reliability and agreement were assessed using intra-class correlations (ICC) and Bland-Altman analysis. Repeatability was tested by estimating the variance of means (ANOVA) of three injections in each patient at different doses. Reproducibility was tested by the ICC of the two time moments. Results: Fifteen patients with heart failure were included. The mean PTT was 11.8 ± 3.1 s at time moment 1 and 11.7 ± 2.9 s at time moment 2. The inter-rater reliability for PTT was excellent (ICC = 0.94). The intra-rater reliability per rater was between 0.81–0.99. Bland-Altman analysis revealed a bias of 0.10 s within the rater groups. Reproducibility for PTT showed an ICC = 0.94 between the two time moments. ANOVA showed no significant difference between the means of the three different doses F = 0.048 (P = 0.95). The mean and standard deviation for PTT estimates at three different doses was 11.6 ± 3.3 s. Conclusions: PTT estimation using CEUS shows a high inter- and intra-rater reliability, repeatability at three different doses, and reproducibility by ROI drawing. This makes the minimally invasive PTT measurement using contrast echocardiography ready for clinical evaluation in patients with heart failure and for preload estimation. Keywords: Contrast enhanced ultrasound, Echocardiography, Indicator dilution technique, Intra-class correlation, Mean transit time, Pulmonary transit time, Reliability

Background estimation [1–3]. This technique uses transthoracic echo- Pulmonary blood volume quantification by transpulmonary cardiography (TTE) to visualize the transcardiac passage of dilution analysis is an essential part of the hemodynamic an ultrasound contrast-agent (UCA) bolus injected in a evaluation to guide fluid management in anesthesia and peripheral vein. Indicator dilution curves (IDCs) are then intensive care practice. Recently, contrast-enhanced ultra- derived from the acoustic backscatter of the UCA bolus in sound (CEUS) has been proposed as a minimally-invasive, the four heart chambers. The mean transit time (MTT) of alternative method for pulmonary transit time (PTT) these acoustic IDCs can be estimated by different methods. The most frequently used methods in clinical practice are based on assessment of the “peaks” of the IDCs or “frame * Correspondence: [email protected] 1Department of Anesthesiology and Intensive-Care, Catharina Hospital counting” of the appearance of the first bubbles in the Eindhoven, Michelangelolaan 2, 5623 EJ Eindhoven, The Netherlands heart chambers [3–5].WeestimatetheMTTbymodel Full list of author information is available at the end of the article

© 2015 Herold et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 2 of 9

fitting using the local density random walk (LDRW) model, positioned in left lateral position. The UCA was admin- which takes into account the Brownian motion of the bolus istered according to our hospital CEUS protocol. All contrast in the blood stream through the pulmonary contrast-enhanced TTE imaging was performed by an vessels and heart chambers [1, 2, 6, 7]. experienced imaging-cardiologist (PH) using an iE33 In previous studies, we demonstrated that volume ultrasound scanner equipped with a S5-1 transducer estimation by CEUS, resulting from the multiplication of (Philips Healthcare, Andover, MA, USA). Four chamber the flow by the PTT (i.e. the difference between the apical views were obtained using harmonic imaging at MTTs of the left atrium and the right ventricle (RV)), 1.3–2.6 MHz, a low mechanical index (MI) of 0.19 to showed excellent agreement with the actual volumes, reduce microbubble destruction, a frame rate of 23 Hz, both in-vitro and in-vivo [2, 6]. Moreover, it showed and a dynamic range of 50 dB with linear post- even better accuracy than transpulmonary thermodilu- processing. tion volume estimation [1, 8]. However, the reliability of SonoVue® (Bracco SpA, Milan, Italy) consisting of PTTs derived with CEUS and the LDRW in-vivo has not microbubbles with a SF6 gas enclosed in a phospholipids been established. Therefore, in this study we investigated monolayer shell was used as UCA. These microbubbles, the reliability and reproducibility of the assessment of PTT with an average size of 3 to 9 μm, pass the pulmonary with CEUS using the LDRW model, in patients referred circulation and can be visualized in both RV and LV [9]. for cardiac resynchronization therapy. We also investi- After a peripheral bolus injection, the passage of micro- gated the effect of different UCA doses on the PTT bubbles through the right and left heart was visualized measurement. In addition, to evaluate the complexity of from a four chamber apical view without breath-hold by the PTT assessment by means of CEUS-recording selecting the proper probe angle, to minimize movement analysis, PTT was also estimated by non-physicians. If artifacts. We diluted 1 ml SonoVue® in saline (1:400, also non-physicians can obtain reliable measurements, 1:200, and 1:100) and injected a bolus of 10 ml of the this would imply a fast learning curve, favoring the solution. Three different CEUS loops were recorded in method adoption in clinical practice. Therefore, our consistent consecutive order. The injections were per- second objective was to evaluate the reliability and formed in a reproducible way as a manual instantaneous agreement between PTT measurements obtained by bolus and each dose was administered as soon as the physicians and non-physicians. recirculating microbubbles disappeared from the left ventricle. These are doses lower than those used for left Methods ventricular opacification [10], but give good opacification Patients of the ventricles while providing an approximately linear As per local hospital protocol, all patients referred for relationship between the UCA concentration and the cardiac resynchronization therapy underwent extensive measured acoustic intensity; the latter is a prerequisite echocardiographic evaluation including contrast enhanced for application of the indicator dilution theory [1, 2]. ejection fraction measurements. This patient population Reliability was tested by eight raters: four physicians scheduled for contrast enhanced TTE to assess ejection who were all experienced with echocardiography but fraction and eligibility were included for this observational with different levels of experience with measuring MTT study. In general, these were patients with symptomatic (one rater was experienced (IH), one rater had some heart failure, a decreased ejection fraction, and QRS- experience (HK), and two had experience in endocardial widening by more than 120 ms. Patients were excluded in border tracing (PH, JD)), and four technicians from the case of atrial fibrillation, an acute coronary syndrome Eindhoven University of Technology of whom three within the past three months, a known allergy to sulphur- were inexperienced with echocardiography and MTT hexafluoride, or a poor acoustic window (impossibility to (HA, GW, AG) and one was an expert in the field visualize an apical 4-chamber view). The Institutional (MM). All raters received a detailed guide for drawing Review Board of the Catharina Hospital Eindhoven ROIs in both RV and LV at the two time moments. Each approved the study, and written informed consent for use rater received the same fifteen anonymized ultrasound of echocardiography data for scientific purposes was apical four chamber view loops following the injection obtained from all subjects. of a SonoVue® in saline bolus at a dilution of 1:200 (Fig. 1). The raters were instructed to draw regions of Measurement protocol interest (ROIs) independently within the endocardial bor- Patients referred for a left ventricle (LV) dyssynchrony ders of the RV and LV throughout the cardiac cycle; no evaluation received a standard of care CEUS echocardi- specified ROI surface was predefined (Additional file 1). ography according to our hospital protocol (Fig. 1). In ROI position was fixed over all the frames. Movement of all patients, an 18-gauge catheter was inserted in a borders or valve structures within the ROI will create arti- peripheral vein of the fore-arm and the patient was facts in the IDC. Therefore, the instructions emphasized Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 3 of 9

Fig. 1 The passage of a bolus of SonoVue® through the right atrium and ventricle (panel b) and left atrium and ventricle (panel c). ROIs are drawn in the right and left ventricle (not shown, see Additional file 1). The acoustic intensities according to panel a, b, and c are expressed in panel d. The acoustic dilution curves of the right (blue indicator dilution curve (IDC)) and left ventricle (red IDC) are fitted according to the local density random walk model and the mean transit times (vertical black lines) are then calculated that the ROIs should be kept within the endocardial The IDCs were fitted by the LDRW model by an inde- border during the entire cardiac cycle. To this end, the pendent researcher (GW). The fitting was performed freeform splines ROIs were drawn using QLab® 8 software automatically as described by Mischi et al. [11]. The ana- (Philips, Healthcare, Andover, MA, USA). The acoustic lysis of the acoustic IDC provides parameters related to intensity over time in the ROIs was expressed as an IDC. the convection and dispersion of the injected SonoVue® To test the reproducibility of the measurement itself, the bolus including the MTT [2, 12]. The difference in MTT raters repeated the drawings in the same dataset within a (ΔMTT) between LV and RV, presenting the PTT, was week interval; these time points are referred to as time derived by subtraction [7]. moment 1 and 2. The raters were blinded for their previous results and to the measurement outcome; also the Statistical analysis order of the patients in the dataset was changed between The analysis and rendering of this observational study time moment 1 and 2. The extracted data were saved as are in line with the guideline for reporting reliability and Excel files and were afterwards analyzed using a custom agreement studies (GRRAS) [13]. The intra-class correl- software to fit the local density random walk (LDRW) ation for random effects models based on repeated- model to the measured IDCs; the method was imple- measures ANOVA was used to evaluate intra-rater and mented in MATLAB® 2009b (The Mathworks, Natick, inter-rater reliability [14]. These were analyzed for the MA, USA) [11]. The interdose repeatability was tested by different time moments 1 and 2 and the different classes: one rater (IH) who drew ROIs in all three different echo physicians versus technicians. The level of agreement loops at the three different doses for the fifteen patients. between the PTT measured by physicians and technicians Thereby, ROI size was measured and the difference in was determined with Bland-Altman analysis to estimate MTT in the fifteen cine loops at three different doses. the feasibility of CEUS using MedCalc Statistical Software Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 4 of 9

version 14.8.1 (MedCalc Software bvba, Ostend, Belgium) Table 1 Demographic characteristics [15]. The distributions of the MTTs of both ventricles and Mean ± SD (n = 15) Min – max (n = 15) the PTTs were tested by the Shapiro-Wilk test. The MTTs Gender (n male/women) 12/3 and PTTs are presented as mean and standard deviation Age (yr) 67 ± 7 58–78 (for normal distributions) or median and interquartile Weight (kg) 81 ± 15 51–109 ranges (for non-normal distributions). Coefficients of – variation were measured for the PTT and computed Height (cm) 176 ± 9 158 188 among the rater groups. The repeatability of the three dif- BMI 26 ± 4 19–37 ferent doses and different loops was tested using univariate BSA 2.0 ± 0.2 1.5–2.3 ANOVA analysis. The ROI size was tested for its distribu- NYHA classification n (%) tion as described above for PTT. Statistical analyses were Class II 7 (47) performed using IBM SPSS statistics for Windows version Class III 5 (33) 22.0 (IBM©, Armonk, NY, USA). Class IV 3 (20) Results Echocardiography Fifteen patients (12 men and 3 women) were enrolled in LVEF (%) 31 ± 11 17–53 the study. Mean age was 67 ± 7 years with an ejection RV dysfunction fraction of 31 ± 11 % (Table 1). The transit times at any TAPSE < 16 mm (n) 4/15 time moment had a normal distribution according to Pulmonary hypertension the Shapiro-Wilk test. The mean MTT of the RV was 8.5 ± 3.1 s and 8.7 ± 3.1 s at time moment 1 and 2, TR velocity > 2.8 m/s (n) 2/15 respectively. For the LV, these values were 20.4 ± 5.6 s Electrocardiogram and 20.3 ± 5.5 s, respectively. The mean PTT was Heart rate 72 ± 17 50–110 11.8 ± 3.1 s at time moment 1 and 11.7 ± 2.9 s at time QRS duration (ms) 151 ± 27 118–194 moment 2. IVMD (ms) in 13 patients 42 ± 14 20–60 The MTT assessments made at the two different time SPWMD (ms) in 11 patients 152 ± 70 10–240 moments per rater showed a high reliability with intra- class correlation coefficient (ICC) for the PTT equal to Comorbidities n (%) 0.94 (95 % CI, 0.90–0.97). For the MTTs of the RV and Hypertension 5 (33) LV, the ICC was 0.98 (95 % CI, 0.97–0.99) and 0.99 Coronary artery disease 11 (73) (95 % CI, 0.98–0.99), respectively (Table 2 and Fig. 2). In Congestive heart failure 14 (93) Fig. 2, the peak of the IDCs of the different ROIs, drawn COPD 6 (40) by the raters, is visualized and its effect on the MTT of Diabetes mellitus 3 (20) the LV evidenced. In this figure, it is shown that the difference in LV MTT is very low among the raters. The Medication n (%) coefficient of variation was lowest for the LV (1.21 %) and Beta-blocker 12 (80) highest for the PTT (3.30 %) (Table 2). Reproducibility for ACE inhibitors 11 (73) the measurements between the two time moments AT II blockers 2 (13) performed by all raters demonstrated an ICC of 0.99 Loop diuretics 12 (80) (95 % CI, 0.98–0.99), 0.99 (95 % CI, 0.98–0.99), and 0.94 K-sparing agents 6 (40) (95 % CI, 0.92–0.94) for the RV MTT, LV MTT, and PTT, respectively (Table 3). Reproducibility between the two Statins 14 (93) time moments for the PTT per rater showed an ICC be- Laboratory tween 0.81 and 0.99. The ICC was 0.99 for all technicians; Creatinine (μmol/L) 105 ± 28 31–160 for the physicians it varied between 0.81 and 0.99 (Table 4). NT-proBNP (pmol/L) 162 ± 111 36–440 Bland-Altman analysis revealed a mean difference of 0.10 SD standard deviation, BMI body mass index, BSA body surface area, NYHA New (±0.54) s between the physicians and technicians. The York Heart Association, LVEF left ventricular ejection fraction, RV right ventricle, 95 % limits of agreement ranged from – 0.95 s to 1.16 s TAPSE tricuspid annular plane systolic excursion, TR tricuspid regurgitation, IVMD interventricular mechanical delay, SPWMD septal to posterior wall motion delay, (Fig. 3). COPD chronic obstructive pulmonary disease, ACE angiotensin-converting ANOVA analysis of the three repeated injections enzyme, AT II angiotensine-II-receptor antagonist, K-sparing potassium sparing, NT-proBNP N-terminal pro-B-type natriuretic peptide showed a mean PTT of 11.4 ± 3.4 s, 11.6 ± 3.4 s, and 11.8 ± 3.2 s at 1:100, 1:200, and 1:400 dilutions of Sono- The measure of effect for the SonoVue® dose on the PTT Vue® in saline, respectively. The variability amongst the accounted for 2 %, η2 0.02. The means and standard means was F 0.048, which was not significant (P =0.95). deviations of the different PTTs per patient per dose are Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 5 of 9

Table 2 Inter-rater reliability of the mean transit times measured volumes measured in an in-vitro model [1]. In this previ- in the right and left ventricle and of the pulmonary transit time of ous study, the volumes were calculated by multiplying the two time moments between the eight raters. The coefficient the flow through the circuit by the difference in MTT of variation is expressed as a percentage between an inflow and outflow tube [1]. A high ICC of ICC (95 % CI) Coefficient of 0.99 (95 % CI, 0.98–1.00) was measured for the three variation (± SD) % repetitions of an UCA bolus at the same flow and – Right ventricle mean transit time 0.98 (0.97 0.99) 2.61 (3.00) volume of the circuit. Left ventricle mean transit time 0.99 (0.98–0.99) 1.21 (1.18) We previously showed that transthoracic and trans- Pulmonary transit time 0.94 (0.90–0.97) 3.30 (3.35) esophageal CEUS can be used to estimate transit times ICC intra-class correlation coefficient, CI confidence interval, SD standard deviation and pulmonary blood volumes in patients [2, 8]. The values of the PTTs derived with CEUS are in line with shown in Fig. 4. The mean PTT and standard deviation cardiopulmonary transit times measured by low-dose for the three different doses was 11.6 ± 3.3 s. The coeffi- contrast-enhanced time-resolved magnetic resonance cient of variation was 5.3 ± 4.4 %. The ROI sizes were (MR) angiography, where the magnetic resonance im- normally distributed and the mean ROI size for the RV aging (MRI) signal is measured in the pulmonary artery was 12 ± 3 mm2 and for the LV 25 ± 6 mm2.The and ascending aorta [16]. In a patient population similar coefficient of variation of the ROI size based on the to ours, with reduced LV function, Shors et al. (2003) three different doses was for the RV 11.3 ± 4.5 % and reported cardiopulmonary transit times of 11.2 ± 4.0 s in 9.0 ± 5.3 % for the LV. patients with a left ventricular ejection fraction (LVEF) between 21 and 30 % and 9.0 ± 1.7 s for a LVEF between Discussion 31 % and 40 % [16]. Our patient population had a mean This study demonstrated that measurement of PTT LVEF of 31 % and the average PTT was 11.8 ± 3.0 s. derived from IDCs of an UCA bolus injected in a per- However, in our cohort two patients had moderate ipheral vein is feasible and highly reliable. The inter- and pulmonary hypertension and four patients had moderate intra-rater correlations are high, and a very low bias right ventricle dysfunction (Table 1), both lead to a between physicians and technicians was observed. prolonged cardiopulmonary circulation times as expressed Therefore, the drawing of the ROI in the CEUS record- by PTT [3]. Although in the study of Shors et al., the ROIs ings is not only reproducible, but also operator inde- were drawn at different regions in the heart and vessels, pendent for measurement of the PTT. According to the the transit times are in line with each other. In this study, high coefficient of variation of the ROI size, the size three-dimensional gradient-echo fast low-angle shots does not influence the PTT measurement. This makes imaging was performed through the pulmonary artery and measurement of the PTT feasible. We also showed that aorta during inspiration breath-holding [16]. This is PTT measurement is repeatable, at three different doses different from our study, where the measurements were in the linear range between concentration and acoustic performed without breath-hold. The advantage of ultra- intensity. The dose did not show any effect on the PTT sound is that the temporal resolution of ultrasound is measurement. The results are in accordance with the much higher (frame rate 23 Hz) than that of MRI

Fig. 2 The fitted curves by eight raters of one patient’s left ventricle indicator dilution curve (IDC) (panel a). The difference in mean transit times of these fitted IDCs among the eight raters in one patient (panel b) Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 6 of 9

Table 3 Intra-rater reliability of the mean transit times measured in the right and left ventricle and of the pulmonary transit time by all the raters between the two time moments ICC (95 % CI) Right ventricle mean transit time 0.99 (0.98–0.99) Left ventricle mean transit time 0.99 (0.98–0.99) Pulmonary transit time 0.94 (0.92–0.96) ICC intra-class correlation coefficient, CI confidence interval

(typically 1 Hz, electrocardiographically triggered on the R-peak depending on the heart rate) [16]. The reliability of our study is in agreement with a recently-published report on the PTT measured in cats using CEUS [17]. The time elapsed to pass the pulmonary circulation by a bolus of SonoVue® was derived by appear- Fig. 3 Bland-Altman analysis. The X-axis represents the mean of the ance time recording in both the pulmonary artery and left average pulmonary transit time (PTT) in the two time moments atrium. Veterinarians with different levels of experience measured by physicians and technicians. The Y-axis represents the evaluated the contrast-enhanced short axis echocardiog- difference of the average PTTs in the two time moments between the physicians and technicians. The solid line is the mean difference raphy loops. Similarly as in our study, no influence of the (bias); dotted lines are limits of agreement [bias ± (1.96 SD)] observers’ experience on PTT measurements was found (median inter-observer variability of 6.8 %) [17]. They concluded that this procedure is simple and robust, and opacification of both ventricles in 27 patients [4]. Incorp- does not need to be limited to experienced operators in orating a model like the LDRW, in the MTT assessment, echocardiography [17]. Notably, our data suggest the makes it less subjective to human errors. The model-free measurement variability to be even lower by the high ICCs parameters do not take into account the underlying (Table 2). This can be explained by the way the individual kinetics of the UCA bolus in the blood stream, like the pulmonaryarteryorchamberMTTswereestimated.In Brownian motion described by the LDRW model [12, 18]. this study, PTT was estimated by the appearance of Though the transit time estimated by two observers at contrast behind the pulmonic valve and in the left atrium. two time moments showed a high intra-observer correl- Appearance of contrast started or stopped a timer [17]. A ation (1 month interval) of 0.92, the inter-observer correl- similar way to estimate PTT was used by Choi et al. [4]. ation was lower (0.79) [4]. This can be explained by the In this study, investigating transit time as an estimate for analyses of the UCA passage, by the timing of the first cardiac output between the RV and LV, the transit time bubble appearance to full opacification of the ventricles; was 3.2 ± 1.2 s in patients with a mean LVEF of 50 ± 16 % this problem could be solved by using LDRW model [4]. The transit time was calculated by identification of the fitting, as shown in our results with a higher inter-rater first bubble appearance, full opacification, and peak

Table 4 Intra-rater reliability per rater for the pulmonary transit time (PTT). Raters 1,2,3, and 8 are physicians and raters 4,5,6, and 7 are technicians. The mean PTTs of the fifteen patients are expressed with their standard deviation at time moment 1 and 2, measured by each rater ICC (95 % CI) Mean PTT t1 (± SD) Mean PTT t2 (± SD) Rater 1 0.91 (0.76–0.97) 11.86 (3.14) 12.13 (3.14) Rater 2 0.99 (0.99–1.00) 11.77 (3.00) 11.69 (2.99) Rater 3 0.92 (0.79–0.97) 11.82 (3.08) 11.49 (3.22) Rater 4 0.99 (0.99–1.00) 11.64 (2.89) 11.70 (2.96) Rater 5 0.99 (0.98–1.00) 11.84 (3.09) 11.82 (2.99) Rater 6 0.99 (0.99–1.00) 11.76 (3.12) 11.71 (3.03) Rater 7 0.99 (0.98–1.00) 11.65 (2.95) 11.60 (2.93) Fig. 4 Means of pulmonary transit times in seconds for each patient Rater 8 0.81 (0.54–0.93) 12.17 (4.15) 11.60 (2.96) based on three different SonoVue® concentrations, expressed as bullets ICC intra-class correlation coefficient, CI confidence interval, SD standard deviation, and standard deviations as error bars t1 time moment 1, t2 time moment 2 Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 7 of 9

reliability. In the feasibility study of the PTT measure- of a left-to-right shunt in patients with a ventricular ment, an ICC of 0.94 was found in nine patients and two septum defect the tail of the IDC will show an extra observers using frame counting [3]. This emphasizes the humb as the indicator will be recycled from the left to simplicity of this procedure, which by using a model- the right atrium [23]. In a right-to-left shunt the IDC based method may become more accurate. will show a biphasic humb at the ascending part of the The total dosage of SonoVue® used to perform the IDC, meaning an earlier increase and decrease of the complete examination according to the protocol was curve followed by the actual peak of a normal passage of 0.875 mg, which is less than the recommended dose by the indicator [23, 24]. In our patients no atrial or Bracco. We did not encounter any side effects of SonoVue®. ventricular septum defects were present. However, given Blood pressure, electrocardiogram, and heart rate were the high prevalence of patent foramen ovale (PFO) of monitored during the examination, and the intravenous 35 % in the general population, we cannot exclude the access was left in place for 30 min. The chance of side presence of a PFO in a part of our patients [25]. Still, a effects has been shown to be low; in a large study with PFO would probably not influence the results as only a 23,188 abdominal CEUS procedures the overall reporting minority of PFO’s will demonstrate a spontaneous right- rate of serious adverse events was 0.0086 % [19]. The to-left shunt and all patients were spontaneous breathing chance of any serious allergic reactions has been shown to without respiratory distress, which means no right-to- have a very low incidence (estimated to be 1:10 000) [10]. left shunts would be expected. Indeed, most PFO’s only We used for the analyses three injections of SonoVue® exhibit shunting during Valsalva manoeuvre. However, at a total amount of 0.875 mg SonoVue®. Thus, we kept knowing that CEUS is contraindicated in patients with all our doses in the linear range between concentration intra-cardiac shunts and the indicator should only pass of SonoVue® and acoustic intensity measured by the the detection point once, the effect of an intra-cardiac ultrasound scanner, avoiding attenuation and shadowing shunt on the MTT of the left ventricle could be of inter- effects [1, 2]. It has been shown that, provided that no est for future studies [10, 26]. shadowing occurs, the effect of the ultrasound scanner Our study population consisted of patients with a settings on the errors in linearization of the video data dilated LV and low ejection fraction; this could benefit exported in DICOM files are kept to a minimum when the feasibility of ROI drawing in the LV. A larger LV the dynamic range is exceeding approximately 45 dB with diminished contractions could facilitate ROI draw- [20]. Then a wide range of gain values can be used with ing, reducing interference with the septal and lateral excellent agreement with those derived from raw radio- walls. Indeed a higher coefficient of variation of the frequency data [20]. The percent of error on the MTT MTT of the much smaller, good contracting RV, 2.61 % measurements, derived from patients with colorectal versus 1.21 % for the LV is supported by the higher coef- liver metastasis, at 50 dB and a gain value of 50 dB was ficient of variation of the ROI size of the RV. Neverthe- 2.8 ± 3.1 % [21]. At lower dynamic range settings, these less, the ICC of 0.98 for the RV is comparable with the error percentages were higher [21]. LV ICC (Table 2). This observation suggests that contribution of ventricular size to the variability of MTT Study Limitation measurement is limited and implies that MTT measure- Although promising, some limitations should be consid- ment in the normal LV is as reliable. ered in the interpretation of our results. Sensitivity of The difference in reliability between the physicians the PTT measurement was not addressed in this study. and non-physicians could not be explained as we did not However, several authors reported over a range of include other performance parameters in the present cardiac function good correlation with the measured study, such as time to complete measurements per ROI, PTT by CEUS. A cut-off point of approximately 4.5 s per patient, for the whole set of patients, and the two distinghuished normals from patients with diminished time moments. The explanation for this difference needs function [3, 4]. In our population, with diminished ejec- to be investigated. tion fractions (± 30 % and dyssynchrony), PTT was Although, an ICC larger than 0.8 is still almost perfect 11.2 s. A good differentiation can be appreciated be- according to Landis and Koch (1977), the intra-rater tween the different ventricular functions based on PTT- reliability per rater showed some variation with ICCs measurements, which is also reported in MRI studies as ranging from 0.81 to 0.99 (Table 4) [27]. This variation described above [16, 22]. can be explained by a deviating MTT measurement at It is known that intra-cardiac shunts due to for ex- one of the time moments, which can probably be over- ample atrial septum defect or ventricular septum defect come by a higher number of repetitions to increase the will influence the IDC [23]. These shunt characteristics precision [28]. In analogy to cardiac output measure- are well known in transpulmonary thermodilution where ments using intermittent bolus thermodilution, three re- a thermistor is positioned in the femoral artery. In case peated injections could be necessary to estimate transit Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 8 of 9

times with a high precision. However, for cardiac out- Authors’ contributions put estimations, the use of four indicator injections IH assisted with the injection of contrast during echocardiography, wrote the ‘ ’ manuscript and performed the statistical analysis. SS performed the analysis, improved the precision to 5 % compared to the true revised the manuscript, implemented the fitting algorithm and modeled the value [28]. The number of repeated injections re- effect of ROI size on the MTT. RAB revised the manuscript and advised on quired to ensure a high precision for MTT estimation statistical analysis. PH performed all the echocardiograms and revised the manuscript. HvA revised the manuscript. MM supervised the fitting algorithm needs to be explored. GUI development and revised the manuscript. EK initiated the trial and In this study, we did not investigate the effect of developed the research protocol including the execution. All authors read different ultrasonographers on the estimation of the and approved the final manuscript. PTT. However, we showed in an earlier study using transesophageal echocardiography that the MTT Acknowledgements We thank Geraldi Wahyulaksana for his contribution in drawing ROIs and measurement is easy to perform; a good and stable fitting all the acoustic intensity curves with MATLAB® 2009b. Alfredo Guillem, view on the chamber or vessel under investigation who performed as a rater. We thank Kathinka Peels and Jan-Melle van Dantzig during the whole cardiac cycle is necessary [8]. It is for their contribution. of importance to keep in- and expirations in a normal Author details and regular pattern, otherwise artifacts may occur due 1Department of Anesthesiology and Intensive-Care, Catharina Hospital to the displacement of the heart in the imaging Eindhoven, Michelangelolaan 2, 5623 EJ Eindhoven, The Netherlands. 2Department of Electrical Engineering, Signal Processing Systems, Eindhoven window. University of Technology, De Zaale, 5612 AZ Eindhoven, The Netherlands. The development of automated algorithms for ROI 3Department of Cardiology, Catharina Hospital Eindhoven, Michelangelolaan definition and MTT estimation could enhance clinical 2, 5623 EJ Eindhoven, The Netherlands. feasibility of this novel CEUS tool [29]. This could Received: 25 October 2015 Accepted: 22 December 2015 create the opportunity to simplify PTT calculation at the bedside. The relationship between the PTT and cardiac References function has been investigated mainly by MRI and 1. Herold IH, Russo G, Mischi M, Houthuizen P, Saidov T, van Het Veer M, et al. radionuclides [16, 30]. The relationship to different Volume quantification by contrast-enhanced ultrasound: an in-vitro comparison with true volumes and thermodilution. Cardiovasc Ultrasound. echocardiographic parameters has recently been investi- 2013;11:36. gated and seems promising, further investigations are 2. Mischi M, Kalker TA, Korsten EH. Contrast echocardiography for pulmonary necessary to evaluate its diagnostic characteristics [3]. blood volume quantification. IEEE Trans Ultrason Ferroelectr Freq Control. 2004;51:1137–47. 3. Brittain EL, Doss LN, Saliba L, Irani W, Byrd 3rd BF, Monahan K. Feasibility and diagnostic potential of pulmonary transit time measurement by contrast Conclusions echocardiography: a pilot study. Echocardiography. 2015;32:1564–71. PTT assessment by drawing ROIs in CEUS recordings is 4. Choi BG, Sanai R, Yang B, Young HA, Mazhari R, Reiner JS, et al. Estimation a reliable technique with a high inter- and intra-rater re- of cardiac output and pulmonary vascular resistance by contrast echocardiography transit time measurement: a prospective pilot study. liability and reproducibility. Each measurement also Cardiovasc Ultrasound. 2014;12:44. showed a high repeatability between three different echo 5. Ugander M, Kanski M, Engblom H, Gotberg M, Olivecrona GK, Erlinge D, loops at three different doses. Differences in ROI size et al. Pulmonary blood volume variation decreases after myocardial infarction in pigs: a quantitative and noninvasive MR imaging measure of hardly affect the MTT per ROI. This makes this novel heart failure. Radiology. 2010;256:415–23. bedside applicable technique for measuring the PTT re- 6. Korsten HH, Mischi M, Grouls RJ, Jansen A, van Dantzig JM, Peels K. liable to be performed by experienced and inexperienced Quantification in echocardiography. Semin Cardiothorac Vasc Anesth. 2006;10:57–62. operators, having an ultrasound scanner and an UCA 7. Mischi M, Jansen AH, Korsten HH. Identification of cardiovascular dilution with intravenous access. This motivates for further systems by contrast ultrasound. Ultrasound Med Biol. 2007;33:439–51. investigation of its clinical application in order to replace 8. Herold IH, Soliman Hamad MA, van Assen HC, Bouwman RA, Korsten HH, Mischi M. Pulmonary blood volume measured by contrast enhanced invasive measurements requiring catheterization. ultrasound: a comparison with transpulmonary thermodilution. Br J Anaesth. 2015;115:53–60. 9. Gorce JM, Arditi M, Schneider M. Influence of bubble size distribution on Additional file the echogenicity of ultrasound contrast agents: a study of SonoVue. Invest Radiol. 2000;35:661–71. 10. Senior R, Becher H, Monaghan M, Agati L, Zamorano J, Vanoverschelde JL, Additional file 1: Apical four chamber view by contrast enhanced et al. Contrast echocardiography: evidence-based recommendations by echocardiography. The SonoVue® bolus passage through the right European Association of Echocardiography. Eur J Echocardiogr. ventricle and left ventricle. ROIs were drawn in the right (red) and left 2009;10:194–212. ventricle (yellow) by one of the raters using Qlab®8 software. The 11. Mischi M, Kalker T, Korsten HHM. Videodensitometric methods for cardiac acoustic time intensity curves are expressed in the bottom panel and output measurements. EURASIP J Appl Signal Processing. 2003;5:479–89. both in the same color as the ROIs. (MOV 5015 kb) 12. Sheppard CW, Savage LJ. The random walk problem in relation to the physiology of circulatory mixing. Phys Rev. 1951;83:489–90. 13. Kottner J, Audige L, Brorson S, Donner A, Gajewski BJ, Hrobjartsson A, et al. Competing interests Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were The authors declare that they have no competing interests. proposed. J Clin Epidemiol. 2011;64:96–106. Herold et al. Cardiovascular Ultrasound (2016) 14:1 Page 9 of 9

14. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420–8. 15. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–10. 16. Shors SM, Cotts WG, Pavlovic-Surjancev B, Francois CJ, Gheorghiade M, Finn JP. Heart failure: evaluation of cardiopulmonary transit times with time- resolved MR angiography. Radiology. 2003;229:743–8. 17. Streitberger A, Hocke V, Modler P. Measurement of pulmonary transit time in healthy cats by use of ultrasound contrast media “Sonovue(R)”: feasibility, reproducibility, and values in 42 cats. J Vet Cardiol. 2013;15:181–7. 18. Wise ME. Tracer dilution curves in cardiology and random walk and lognormal distributions. Acta Physiol Pharmacol Neerl. 1966;14:175–204. 19. Piscaglia F, Bolondi L. The safety of Sonovue in abdominal applications: retrospective analysis of 23188 investigations. Ultrasound Med Biol. 2006;32:1369–75. 20. Rognin NG, Frinking P, Costa M, Arditi M. In-vivo perfusion quantification by contrast ultrasound: Validation of the use of linearized video data vs. raw RF data. Ultrasonics Symposium, 2008 IUS Proceedings, IEEE 2008:1690-3. 21. Gauthier TP, Averkiou MA, Leen EL. Perfusion quantification using dynamic contrast-enhanced ultrasound: the impact of dynamic range and gain on time-intensity curves. Ultrasonics. 2011;51:102–6. 22. Cao JJ, Wang Y, McLaughlin J, Haag E, Rhee P, Passick M, et al. Left ventricular filling pressure assessment using left atrial transit time by cardiac magnetic resonance imaging. Circ Cardiovasc Imaging. 2011;4:130–8. 23. Giraud R, Siegenthaler N, Park C, Beutler S, Bendjelid K. Transpulmonary thermodilution curves for detection of shunt. Intensive Care Med. 2010;36:1083–6. 24. Michard F, Alaya S, Medkour F. Monitoring right-to-left intracardiac shunt in acute respiratory distress syndrome. Crit Care Med. 2004;32:308–9. 25. Meier B, Lock JE. Contemporary management of patent foramen ovale. Circulation. 2003;107:5–9. 26. Zierler KL. Theoretical basis of indicator-dilution methods for measuring flow and volume. Circ Res. 1962;10:393–407. 27. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–74. 28. Nilsson LB, Nilsson JC, Skovgaard LT, Berthelsen PG. Thermodilution cardiac output–are three injections enough? Acta Anaesthesiol Scand. 2004;48:1322–7. 29. Saporito S, Herold IH, Houthuizen P, van den Bosch HC, Korsten HH, van Assen HC, et al. Automatic indicator dilution curve extraction in dynamic-contrast enhanced imaging using spectral clustering. Phys Med Biol. 2015;60:5225–40. 30. Jones RH, Sabiston Jr DC, Bates BB, Morris JJ, Anderson PA, Goodrich JK. Quantitative radionuclide angiocardiography for determination of chamber to chamber cardiac transit times. Am J Cardiol. 1972;30:855–64.

Submit your next manuscript to BioMed Central and we will help you at every step:

• We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit